Polish Research on Biometrics and Deepfakes at WACV 2026

The event is one of the most important gatherings dedicated to computer vision. Four representatives of the Center for Artificial Intelligence Security Research participated in the conference, presenting four scientific papers.

During the 5th Workshop on Image/Video/Audio Quality Assessment in Computer Vision, VLM and Diffusion Model, they presented the paper “VIBEFACE – Video and Image Biometric Dataset for Evaluation of Faces”, devoted to the biometric database developed by our team. The work was also showcased during the poster session, which enabled numerous discussions with researchers interested in biometrics and image analysis.

First day:

• Dr Ewelina Bartuzi‑Trokielewicz discussed the limitations of existing deepfake detectors, emphasizing that although they perform well on reference datasets, they often fail when confronted with real‑world incidents.
• Donat Stankiewicz presented the team’s audio dataset RaT, highlighting its unique characteristics and advantages over existing resources.
• Alicja Martinek presented an analysis of deepfakes involving the likenesses of politicians and discussed how such materials affect the performance of biometric systems.

Second day:

• Michał Koźbiał presented the VIBEFACE dataset, which includes a wide range of visual data applicable to biometric research.

In addition to their own presentations, the researchers had the opportunity to attend many inspiring talks and engage in numerous discussions with scientists from around the world. We hope that the connections established during the conference will lead to long‑term collaborations and, eventually, joint research projects.

Dataset – https://nask-biometria.github.io/VIBEFACE/